Profibus DP is not broken — millions of nodes run fine. But every new device generation, diagnostic feature and engineering convenience lands on Profinet first, and mixed plants need a deliberate strategy rather than drift.
What genuinely changes
Physics: RS-485 daisy chains with termination discipline give way to switched Ethernet — stars, rings, trees, whatever the plant needs, with standard cables and no termination folklore. Speed and size: 100 Mbit/s and address space stop being constraints; update times configure per device instead of per-bus compromise. Diagnostics: the biggest practical win — per-port statistics, topology views, device-level maintenance states, all visible from the engineering tool instead of a handheld bus analyzer. Names over addresses: devices carry names; replacement units get configured by topology automatically when you design for it.
What does not change: the need for design discipline. Ethernet forgives sloppy wiring less than folklore claims (EMC still exists), and update times still get budgeted, not wished.
Mixed-plant strategy
Keep Profibus segments that are healthy and stable — behind an IE/PB Link proxy they appear as Profinet devices and buy years of transition calm. Route every new installation to Profinet. Migrate old segments when their devices go obsolete anyway; converting a working DP slave estate for ideology burns money.
FAQ
Is Profinet deterministic enough for drives? RT class handles typical drive control comfortably; IRT covers the isochronous motion tier. Determinism is a design property — reserve bandwidth and plan topology.
Do I need managed switches everywhere? On machine networks, Profinet-aware switching (often built into devices) is the default; the plant backbone deserves managed infrastructure with diagnostics.
Zone Otomasyon plans bus migrations that ride the obsolescence wave instead of fighting it. Network modernization.